2. “MACHINE LEARNING TECHNIQUES
FOR THE DETECTION AND
TRANSCRIPTION OF VARIABLE
MESSAGE SIGNS ON ROADS “
NAME :- R.VIKRAM CHAKRAVARTHY
ID.N O:- 18M91A0597
GUIDED BY :- DR.M. Sridhar
4. ABSTRACT
Among the reasons for traffic accidents, distractions are the most common. Although there are many
traffic signs on the road that contribute to safety, variable message signs (VMSs) require special attention,
which is transformed into distraction. ADAS (advanced driver assistance system) devices are advanced
systems that perceive the environment and provide assistance to the driver for his comfort or safety.This
project aims to develop a prototype of aVMS (variable message sign) reading system using machine
learning techniques, which are still not used, especially in this aspect.The assistant consists of two parts: a
first one that recognizes the signal on the street and another one that extracts its text and transforms it
into speech. For the first one, a set of images were labeled in PASCALVOC format by manual annotations,
scraping and data augmentation. With this dataset, theVMS recognition model was trained, a Retina Net
based off of ResNet50 pretrained on the dataset COCO. Firstly, in the reading process, the images were
preprocessed and binarized to achieve the best possible quality. Finally, the extraction was done by the
Tesseract OCR model in its 4.0 version, and the speech was done by the cloud service of IBMWatsonText
to Speech.
5. INTRODUCTION
Since the democratization of the private car, the world’s fleet has continued to grow.This increase has
brought with it the problem of traffic accidents. Data from theWorld Health Organization (WHO) estimate
that during the period 2011–2020, 1.1 million people died due to traffic accidents and between 20 and 50
million were injured .
In Spain, the Dirección General deTráfico (DGT) has produced a series of statistical yearbooks, which
illustrate the evolution from 1960 to 2018 .
Generally speaking, the number of casualties has increased in recent years.The number of fatalities and
hospitalized victims has decreased while the number of non-hospitalized injured victims has increased.
Accidents are still occurring, but the probability of death is decreasing.The causes of traffic accidents can be
classified according to the risk factor that causes them.They are distinguished by human, mechanical and
environmental factors (the state of the asphalt or traffic signs and weather conditions).According to the DGT,
in 2018, 88% of accidents were the result of inappropriate driver behaviors (similar conclusion to study ,
which states that 90% are due to human causes). In first place were distractions (33%), followed by speeding
(29%) and alcohol consumption (26%) .The same organization has prepared a document that lists the main
distractions and explains how they affect accidents . It shows that actions such as using a cell phone, eating or
smoking are activities that require time and attention, reducing concentration while driving.
6. The driver’s physical condition also affects his reaction time and ability to be distracted.This has a direct
impact on braking distance, which is a serious risk. Many of these behaviors are known to drivers and many
are declared offenders .
The WHO, in its report on 2011–20, proposes five action points to improve safety. Examples of the third
(safer vehicles) are initiatives such as Prometheus created by an association of vehicle manufacturers and
researchers, or DRIVE (Dedicated Road Infrastructure forVehicle Safety in Europe), funded by the EU
(European Union) , which has promulgated a large number of papers on fundamental and practical
problems, such as GIDS (Generic Intelligent Driver Support) .
Its aim was to “to determine the requirements and design standards for a class of intelligent co-driver
(GIDS) systems that are maximally consistent with the information requirements and performance
capabilities of the human driver” .
It was the beginning of what we know today as ADASs (advanced driver assistance systems), successors to
basic safety systems and enablers of autonomous driving in the future .Variable message signs (VMSs) are
roadside ATIS (advanced traveler information system) devices consisting of LEDs (light-emitting diodes)
that stand out against a black background .
They are the mechanism used by traffic agencies to communicate useful information to drivers in order to
improve their safety.These messages convey information by means of personalized text and/or traffic sign
pictograms
7. METHODOLOGY
The processing steps are summarized in fig 1.The images captured by the vehicle camera are initially
processed by theVMS object recognition module. the next step is to normalize the section that
corresponds to theVMS by cropping the image ,changing the perspective and angle in addition to
adjusting the colour to facilitate the following task of extracting the text from the image. Finally , the text
is converted to audio using a “text to speech” service in the cloud.
9. INPUT : Video image
VMS Object detection
OUTPUT : VMS Image
Image filtering
Text extraction
Text to speech
THE VMS READING PROCESS CONSISTS OF 2 STEPS :-
These processing steps for theVMS speech system are divided into two subsystems combining local processing and
cloud services : aVMS recognizer and a content extractor and speaker
A : VMS extraction B : Processes the image to extract the content and speak it
PANEL EN PRUEBAS
Fig:-2
10. COMPONENTS
The Lidar system
GPS
Radar Sensors
Ultrasonic Sensors
Central Computer
12. EXISTING
TECHNOLOGY
PHEV Batteries can be charged by an
outside electric power source , an
internal combustion engine , or
regenerative braking.
As an traditional petrol or diesel
engines can be refilled again from
external source as the main and runs
on fuel alone for a unlimited distance.
Fig:-5
13. PROPOSED
TECHNOLOGY
PHEVs use batteries to power an
electric motor and another fuel ,
such as gasoline or diesel , to power
an internal combustion engine or
another propulsion source
The battery can be charged from an
external source such as the mains
and run on battery power alone for a
limited distance.
Fig:6
15. CONCLUSION
We have seen massive changes, particularly in terms of technology, but also in terms of people’s attitude
towards cars’ environmental impacts and other mobility solutions, from the first electric car established in
1837 up to the present time.Although the electric vehicle market is currently a lucrative goal for companies
and start-ups in India, several obstacles still remain to be addressed in order for EVs to be ready for mass
adoption. High-cost barriers include, for example, manufacturing electric vehicles domestically.
Similarly, battery manufacturing is essentially a costly venture.The Indian Government must concentrate its
energies on promoting technological disruption to resolve these challenges.The government would also
need to provide enhanced tax incentives and subsidies to potential car owners and suppliers in order to
quicker adoption of EVs.